What Prediction Markets Do to a Sportsbook's Numbers: 2026 Evidence
Published 2025–2026 volume and revenue figures for prediction markets, what Sportradar, Flutter, DraftKings and Robinhood reported, and a unit-economics table.
TL;DR
Prediction markets produced real, published numbers in 2025 and 2026, and almost all of them belong to two US-facing exchanges and three US-listed operators. What is measured: Kalshi and Polymarket cleared more than 44 billion dollars of notional volume in 2025 and 53 billion dollars in July 2026 alone, about four-fifths of it on sport; Robinhood booked 156 million dollars of event-contract revenue in the second quarter of 2026; Flutter expects about 50 million dollars of market-making revenue in 2026 and calls FanDuel Predicts' quarterly revenue "not material"; DraftKings reports Predictions inside one sports segment; Sportradar's chief executive put its 2026 upside "in the tens of millions" after deals took longer than planned. What is claimed: cannibalisation estimates, vendor pipelines and forward valuations. This post separates the two and ends with a unit-economics table an operator can fill with its own assumptions.
Key takeaways
- Notional volume is not handle: published volumes count every contract at one dollar of notional value, so a contract bought at five cents counts as a dollar and the cash at risk is smaller than the headline.
- Sport is the engine: 80% of Kalshi's volume and 39% of Polymarket's since July 2024, and about four-fifths of both platforms' volume in July 2026.
- The operators that disclose anything disclose revenue, not profit: Robinhood 156 million dollars in Q2 2026; Flutter 6 million dollars of market-making revenue in Q2 and about 50 million expected for 2026; DraftKings does not break Predictions out. Both sportsbooks reported lower sports net revenue margins while investing behind the new products.
- Suppliers see the money arriving later than they said: Sportradar cut its 2026 guidance on 3 August 2026 and expects "tens of millions" from prediction markets in 2026.
- The right model for a European book is a fixed-odds margin model, not an exchange take rate: handle, theoretical margin, realised hold, promotions, vendor share, settlement cost and cross-sell, each labelled fact or assumption.
Volume: what the trackers show
The volume record is kept by The Block, a digital-assets data firm, and the most careful reading of it is Pew Research Center's. The Block wrote on 31 December 2025 that Kalshi and Polymarket, "a de facto duopoly", had generated "more than $44 billion in trading volume this year", with combined monthly volume topping 10 billion dollars in November 2025. On 1 May 2026 it reported that the two platforms' combined lifetime volume had crossed 150 billion dollars in April 2026, that April was the first monthly decline since September 2025 after seven record months, and that Polymarket's active traders fell to about 643,000 in April from more than 733,000 in March.
Pew's analysis of 27 May 2026 put combined monthly global volume at under 5 billion dollars in September 2025 and about 24 billion dollars in April 2026, and offered the comparison that matters to a sportsbook: US legal sportsbooks took about 14 billion dollars a month on average in 2025. Pew's follow-up of 23 September 2026 shows the World Cup effect: combined volume of 25.7 billion dollars in May, 47.7 billion in June and 53.0 billion in July 2026, which sum, from Pew's category table, to Kalshi at 16.8, 33.0 and 40.1 billion and Polymarket at 8.8, 14.7 and 12.9 billion. DLA Piper's tracker of 2 September 2026 cites 51 billion dollars for prediction markets overall in 2025 and a Bernstein Research estimate of 240 billion by the end of 2026; the second is an estimate, not a measurement.
Two definitional notes change what these numbers mean. Pew states that trading volume is "notional taker volume in U.S. dollars" and that each contract is counted at its "$1 notional value" rather than its trade price, so a contract bought at five cents counts as a dollar of volume. Adkuu's reading is that the cash actually staked is well below reported volume whenever prices sit far from fifty cents, which makes a sportsbook's handle, money staked, a smaller unit than the "volume" quoted against it. Scope also differs: Pew's September figures include Polymarket US; its May figures refer to Polymarket International unless stated.
Sport's share
Since July 2024, sport has been 80% of Kalshi's volume and 39% of Polymarket's, with politics at 4% and 32% and cryptocurrency at 7% and 20%, according to Pew's May 2026 analysis; in October and November 2024, before Kalshi added sport, politics was 90% of its volume and 65% of Polymarket's. By July 2026 the two had converged: Pew's September table shows sport at 31.4 billion of Kalshi's 40.1 billion dollars and 10.6 billion of Polymarket's 12.9 billion, about 78% and 82% by Adkuu's arithmetic. For a European book the lesson is that the headline growth is mostly a sports product sold in a different legal wrapper; the non-sport remainder, politics and crypto above all, is the category an operator would actually be adding.
Revenue: what the listed operators filed
Robinhood's second-quarter release of 29 July 2026 is the only filing that isolates the product. Event-contracts revenue was 156 million dollars, "up over 10x" year on year; "Event Contracts Traded increased over 10x year-over-year to a record 13.6 billion"; total net revenues rose 32% to 1.31 billion dollars and transaction-based revenues 44% to 776 million. Each contract trades in one-cent increments up to one dollar and pays one dollar at settlement, so by Adkuu's arithmetic the company earned about 1.1 cents per contract traded. Robinhood launched Rothera, a CFTC-licensed exchange and clearinghouse run as a joint venture with Susquehanna, in June 2026; the release says it has traded over 3.5 billion contracts to date and carries a risk factor that regulatory actions could prevent the company from offering event contracts.
Flutter's release of 5 August 2026 gives the sportsbook view. Chief executive Peter Jackson wrote: "We view prediction markets as a very attractive opportunity, incremental to sports betting and iGaming", that "We continue to see a limited cannibalization impact on our existing customer base", and that "While operational progress in H1 has been slower than planned, we are gaining traction." FanDuel Predicts' second-quarter revenue "was not material"; second-half gross revenues "are expected to be offset by customer acquisition investment costs before increasing in 2027"; market-making produced 6 million dollars in the quarter and "approximately $50m of revenue" is expected for 2026; sports and novelty contracts are moving from CME to Crypto.com. The same release shows the cost of the push: sales and marketing up 61%, promotional spend at 5.4% of handle, and a US net revenue margin of 8.7%, down 170 basis points, against a structural margin of 14.0%.
DraftKings' release of 6 August 2026 does not break Predictions out at all: "Sports Consumer Volume" of 13.1 billion dollars (up 15%) and a "Sports Net Revenue Margin" of 6.8% (from 8.7%) both combine Sportsbook and Prediction Markets, revenue fell 5% to 1,443 million dollars, the net loss was 67.6 million, and chief executive Jason Robins said "Predictions is already growing faster than we anticipated." iGB reported on 7 August 2026 that about 600,000 customers had engaged with the predictions platform since the start of the year.
Neither exchange has filed revenue figures; the only numbers are press reports. CoinDesk's report of 13 August 2026, citing The Information, put Kalshi's annualised revenue at about 4 billion dollars in July 2026 on World Cup trading, sport at over 80% of its volume, and the company in talks to raise at a 40 billion dollar valuation; those are reports, not filings.
Suppliers: what Sportradar, SOFTSWISS and DATA.BET said
Sportradar signed both exchanges in 2026: a multi-year agreement with Kalshi announced on 8 June 2026 covering official data and live odds to support timely settlements, fan engagement, customer acquisition and integrity services across MLB, NHL, MLS and UFC among others; and an expansion with Polymarket on 27 August 2026 to more than 20 leagues and competitions, roughly 300,000 matches a year, with exclusive streaming for the Bundesliga, Euroleague Basketball and the ATP Tour.
The money is slower than the announcements. On 3 August 2026 Sportradar reported second-quarter revenue of 378 million euros, up 19%, and a 4 million euro net loss, and cut its full-year revenue guidance to 1.52 to 1.53 billion euros from 1.56 to 1.58 billion, according to NEXT.io. The earnings-call summary carried by Yahoo Finance quotes chief executive Carsten Koerl: "For 2026, the upside is in the tens of millions" (the company reports in euros), "the ramp-up will be significant in 2027 and 2028", and "2026 was largely spent negotiating deals and convincing league partners"; chief financial officer Craig Felenstein said the deals "include both a fixed fee and a variable fee component to capture upside". Koerl also said clients report "very limited cannibalization" of online sports betting because prediction markets are mainly distributed in states without it, such as California, Texas and Florida.
Two B2B suppliers in Europe have published product facts and pipeline, not operator revenue. SOFTSWISS launched a fixed-odds Prediction Markets product on 2 April 2026, with operators controlling pricing and margins and an iFrame or sportsbook integration in two to three days for existing partners, and said on 29 June 2026 that more than 50 projects had entered its pipeline since launch, two of them live outside its own sportsbook ecosystem. DATA.BET launched Prediction Markets as a standalone vertical on 29 April 2026 (politics, geopolitics, finance, technology, crypto, economy, culture and weather; binary and multi-outcome markets; iFrame at launch) and on 9 September 2026 added crypto Up or Down markets at decimal odds across five settlement horizons, with API access. Neither release states volumes or revenue.
What the evidence does not show
None of the sources opened for this post contains a European operator's published prediction-market handle or revenue; every operator figure above is American. The cannibalisation claims come from Flutter and Sportradar, both of which have a commercial stake in it, and neither published a method; their own numbers show the cost side (marketing up 61% at Flutter, margin down at DraftKings) ahead of the revenue side. The distribution of profit among exchange users is covered in Adkuu's reading of the Wall Street Journal's account-level data in Sharks eat fish. And settlement is a cost line the volume figures hide: Kalshi's own help page says determining an outcome "can take anywhere from one hour to more than twelve hours" after a market closes.
What an operator should model
The exchanges earn a take rate on volume. A European book running fixed odds earns a margin on stakes, so the model is the sportsbook's own, with a few lines that behave differently. Every figure in the "illustrative assumption" column is an assumption, not a fact.
| Driver | What to model | Illustrative assumption | What the 2026 evidence says |
|---|---|---|---|
| Handle | Money staked, not notional volume; share of sportsbook actives who try it; average stake | 2% of sportsbook handle in year one | No European figure published; Pew's US comparison (14 billion dollars a month of sportsbook handle in 2025 against 24 billion of notional in April 2026) is not like for like |
| Theoretical margin | The overround on a two-way market | A 50/50 question at 1.85 and 1.85: a 108.1% book, a margin of 7.5% of stakes | Fixed-odds suppliers leave the margin to the operator (SOFTSWISS; Adkuu Pulse); an exchange earns a take rate (Robinhood: about 1.1 cents per contract, Adkuu's arithmetic) |
| Realised hold | Margin after results variance and bet mix | 6% to 9% of stakes | Flutter US net revenue margin 8.7% and DraftKings sports net revenue margin 6.8% in Q2 2026, both blended with sportsbook |
| Promotions | Bonus and free-bet cost as a share of handle | 2% of handle | Flutter promotional spend 5.4% of handle in Q2 2026; FanDuel Predicts second-half gross revenue offset by acquisition costs |
| Vendor cost | Revenue share on GGR or a fixed fee | 10% of GGR | Adkuu Pulse is revenue share (rate not published); Sportradar's exchange deals are fixed plus variable |
| Settlement cost | Markets per month, share needing a human look, minutes each, voids | 5% of markets, 15 minutes each | Kalshi: one to more than twelve hours per outcome; Adkuu Pulse: automated tiers with human review as fallback |
| Cross-sell | Incremental sportsbook or casino revenue from players acquired through prediction markets | Zero until measured | Flutter and Sportradar say "limited cannibalization" and "incremental"; neither quantified it |
| Gambling tax | GGR tax at the national rate | Not modelled | Varies by country |
A worked example at those assumptions: 1,000,000 euros of monthly stakes at a 7.5% theoretical margin is 75,000 euros; realised hold of 6% to 9% gives 60,000 to 90,000; promotions at 2% of handle take 20,000, leaving 40,000 to 70,000; a 10% vendor share on the gross comes to 6,000 to 9,000, leaving roughly 34,000 to 61,000 euros before tax and operations. If 300 markets settle in the month and 5% need fifteen minutes of a trader's time, settlement costs under four hours. The arithmetic says where to look: promotional cost and hold variance move the answer far more than the vendor share, and the one line nobody has published, cross-sell, is the one that would change the decision.
Where Adkuu Pulse sits
Adkuu Pulse prices every market as fixed odds with the operator's margin: for each question it delivers an AI probability shaped by a risk model (flow-weighting, liability, staleness, circuit breakers), then ready-to-use odds priced with a margin formula the operator controls, over REST, WebSocket and gRPC, with resolution automated in tiers (data API, web search, consensus, human review as fallback) and a revenue-share commercial model. In Adkuu's model the fixed-odds form fits inside an operator's existing sports-betting licence; classification is confirmed per market with counsel. For the table above that means the margin and hold lines are the operator's own numbers and the settlement line is mostly automated. Details are on the Adkuu Pulse page; the regulatory map is in Prediction markets under a betting licence and the vendor questions in the feed buyer's checklist.
FAQ
How big were prediction markets in 2025 and 2026? Kalshi and Polymarket cleared more than 44 billion dollars of notional volume in 2025 (The Block) and 53.0 billion dollars in July 2026 alone (Pew Research Center, on The Block's data). Those figures count each contract at one dollar of notional value, so cash staked is lower.
What share of prediction-market volume is sport? 80% on Kalshi and 39% on Polymarket since July 2024 (Pew, May 2026); by July 2026 roughly four-fifths on both, by Adkuu's arithmetic from Pew's September 2026 table.
Do prediction markets cannibalise a sportsbook? Flutter said in August 2026 that it sees "a limited cannibalization impact", and Sportradar's chief executive said clients report "very limited cannibalization" because the product is sold mainly where online sports betting is not yet regulated. Neither published a method, and no European operator has published data.
What margin can a fixed-odds operator expect on binary markets? Whatever it prices in: a two-way market at 1.85 and 1.85 carries a 7.5% theoretical margin. Realised hold depends on results and bet mix; the only published comparators are blended US sportsbook margins of 6.8% (DraftKings) and 8.7% (Flutter) in Q2 2026.
Sources
- Robinhood Reports Second Quarter 2026 Results — Robinhood Markets, Inc. (Form 8-K exhibit)
- Flutter Entertainment Announces Q2 2026 Financial Results — Flutter Entertainment plc (Form 8-K exhibit)
- DraftKings Reports Second Quarter Results — DraftKings Inc. (Form 8-K exhibit)
- Sportradar and Kalshi Announce First-of-its-Kind Data and Infrastructure Global Partnership for Prediction Markets — Sportradar
- Sportradar and Polymarket Significantly Expand Prediction Market Partnership to More Than 20 Global Sports Leagues and Competitions — Sportradar
- Kalshi and Polymarket trading volumes dramatically increase since mid-2025 — Pew Research Center
- Prediction markets' trading volume doubled between May and July, largely driven by sports — Pew Research Center
- Prediction markets explode in 2025: Inside the Kalshi-Polymarket duopoly and challengers — The Block
- Q2 2026: Sportradar cuts guidance despite 19% revenue rise — NEXT.io
- Sportradar Group AG (SRAD) (Q2 2026) Earnings Call Highlights — Yahoo Finance UK (GuruFocus)
Last verified: October 10, 2026.